For objective 1, blood from healthy dogs was collected into EDTA anticoagulant-containing vacutainer tubes (BD Biosciences, Franklin Lakes, NJ, United States) by jugular or cephalic venipuncture. The healthy dogs were research animals housed at our institution or owned by students, faculty, and staff, and blood was obtained with owner consent. For objective 2, bone marrow was aspirated from the humeri of research beagles housed at our institution. Blood and bone marrow sample collection was approved by the Institutional Animal Care and Use Committee (#2009–0085). For objective 3, we used data from two sources: (1) Blood, aspirates from bone marrow, lymph node or masses, and body cavity fluid samples collected from dogs with hematopoietic neoplasia that were submitted to the AHDC for immunophenotyping as part of routine diagnostic service. These samples and data become the property of the AHDC after submission; (2) Blood, bone marrow or lymph node aspirates collected from dogs, with client consent, as part of an ongoing research study on acute leukemia. Data collection for the study herein ceased on July 6, 2023. Two immunophenotyping panels were used: One for acute leukemia and one for lymphoma. For routine immunophenotyping, the panel was chosen by the requesting clinician, with or without consultation with a clinical pathologist at our institution. Typically, lymphoma panels were requested for lymph node aspirates, acute leukemia panels were requested for bone marrow, mass or body cavity fluid aspirates, and both panels were run on blood samples. The acute leukemia panel was used for the research study. Each panel consisted of antibodies against various markers, with all panels including the anti-CD80 antibody. In cases phenotyped after April 2021, including all cases in the research study, we used triple-marker antibody combinations; before April 2021, the same antibodies were used as single markers with a few double-marker combinations (Tables 1, 2). Additional conjugated and unconjugated antibodies against other antigens were used as single markers in the acute leukemia panel, e.g., CD61-phycoerythrin (Beckman Coulter Cat# IM3605, RRID:AB 131237) and the α and β chain of the T cell receptor (TCRαβ, Peter Moore, clone CA15.8G7).

Because the intended application of the antibody was for flow cytometric-based immunophenotyping, we used flow cytometric analysis to verify the labeling pattern of the anti-CD80 antibody in healthy canine blood in three ways: (1) Double- or triple-labeled analysis of normal dog leukocytes; (2) Flow cytometric-based sorting of anti-CD80 and -CD14 double-labeled leukocytes followed by cytologic analysis of sorted cells, using CD14 as a monocyte marker; and (3) Isolation of leukocytes followed by flow cytometric labeling with the anti-CD80 antibody. Monocytes, T cells and B cells were isolated by immunomagnetic bead-labeling, whereas neutrophils were isolated by double-density centrifugation followed by red blood cell (RBC) lysis. The same antibodies were used for verification of staining pattern of the anti-CD80 antibody and immunophenotyping of clinical cases (Table 1).

For the bone marrow aspirates, we labeled bone marrow mononuclear cells (BMMC) with the anti-CD80 antibody and then performed: (1) Flow cytometric analysis; and (2) Sorting for cells labeled with the anti-CD80 antibody followed by cytologic analysis of the positively and negatively stained sorted cells. All reagents were from Sigma-Aldrich (St Louis, MO, United States), unless otherwise specified.

Flow cytometric panels, including the anti-CD80 antibody, were performed on blood and lymph node, bone marrow, mass or body cavity fluid aspirates from dogs with hematopoietic neoplasia, including lymphoma and leukemia, using BD FACSCalibur^™, Accuri C6 (BD Biosciences) or Novocyte (2000R, Agilent Technologies, Santa Clara, CA, United States) flow cytometers. Various software was used to analyze the data, including BD FACScan software, FloJo^™ and FCS Express (De Novo Software, Dotmatics, Pasadena, CA). In samples collected after July 2019, 7-AAD was used to gate out dead cells, otherwise cell debris was excluded from analysis based on its location in a FSC vs. SSC dot plot. A tumor cell gate was created in the FSC vs. SSC dot plots, based on abundance of events and characteristic location (low to high FSC, low to medium SSC). When possible, residual normal leukocytes (neutrophils, monocytes, lymphocytes) were gated separately in the same plot. Antibody labeling of each gate was assessed in histogram, SSC vs. fluorescence, and quadrant fluorescence plots for triple labels or histograms and quadrant plots for single or double labels. Only data from the region containing the tumor cells was included in this study; however, antigen expression on residual normal leukocytes was assessed as internal positive and negative controls for antibody staining. For all markers other than CD34, expression on ≥20% of gated tumor cells was considered positive, in relation to an isotype control (31, 32); for CD34, ≥5% labeling of the tumor cells was considered positive (2, 3, 33). Since monocytes and lymphocytes can overlap or fall in the tumor cell gate, to avoid including these cells in the analysis of CD80 expression, we compared the percentage of CD80⁺ cells with the percentages of CD14⁺ cells and monocytes in a differential cell count. We also assessed the location of CD80⁺ and CD14⁺ cells in FSC and SSC plots. When the percentage of CD80⁺ and CD14⁺ cells were similar (± 10%) and events overlapped in a typical location for monocytes in a FSC and SSC plot (medium to high FSC and medium SSC), they were considered residual monocytes. Similarly, when the tumor gate included residual normal lymphocytes (mostly in lymph node aspirates), we identified these normal cells based on their low FSC and SSC characteristics and flow cytometric results showing a mixed population of lymphocytes, i.e., CD21⁺/CD22⁺ B cells and CD3⁺/CD4⁺ and CD3⁺/CD8⁺ T cells.

Based on morphologic features and hematologic and flow cytometric results, the hematopoietic neoplasms in the dogs were classified as B or T lymphoma/leukemia, B or T chronic lymphocytic leukemia (CLL), acute myeloid leukemia (AML), or “mixed lineage” leukemia. Classification of lymphoid neoplasms as B or T was based on tumor cell expression of CD21⁺ or CD22⁺ or both and CD3⁺ or CD5⁺ or both, respectively. In select cases, immunocytochemical staining for CD3 was performed on blood or cytology smears to confirm a T cell origin or verify weak or negative flow cytometric reactions for CD3, with ≥20% CD3⁺ tumor cells being defined as a positive reaction. B or T cell lymphoma/leukemias consisted of intermediate to large cells (“blasts”) and were grouped as a single entity, regardless of the blast percentage in blood or bone marrow. Chronic lymphocytic leukemia was characterized by a mature lymphocytosis, with negative test results for tick-borne diseases. For B-CLL, we used previously defined criteria of > 5.0 × 10⁹ small lymphocytes/L with > 60% CD21⁺ cells (34). Similar criteria are lacking for T-CLL in dogs, so we used a cut-off of > 20.0 × 10⁹ lymphocytes/uL with > 60% CD3 or CD5 expression. An acute leukemia was classified as AML on flow cytometric analysis if there were ≥ 20% blasts in blood or bone marrow, lymph node, tissue, or body cavity fluid aspirates and blasts were positive for myeloid-associated antigens CD4, CD11b, CD11c, CD14, and CD18, alone or in combination and negative for B and T cell markers (3). Because myeloid cells (monocytes, neutrophils, and eosinophils) typically comprise <5% of cells in normal lymph node aspirates (33, 35), we defined an extramedullary AML in lymph node aspirates as ≥20% blasts in cytologic smears combined with ≥5% positive reactions for myeloid-associated antigens on flow cytometric analysis. In cases in which we performed triple labeling with CD34/CD80/MHCII, CD34/CD11b/CD11c or CD34/CD11b/CD18, a leukemia was also classified as AML if there were ≥ 5% CD34⁺/CD11b⁺, CD34⁺/CD11c⁺, or CD34⁺/CD18⁺ cells. In acute leukemias that could not be phenotyped by flow cytometry, cytochemical staining was done to distinguish between AML and ALL, as described (2, 3, 36). Expression of alkaline phosphatase (ALP), alpha-naphthyl butyrate esterase (ANBE), chloroacetate esterase (CAE), myeloperoxidase (MPx), and Sudan Black B (SBB) in ≥3% of the cells, alone or in combination, was consistent with a myeloid lineage for an acute leukemia (Supplementary Table S1) (2, 3). AML was further classified into “not otherwise-specified” categories of unclassified, myelomonocytic, monocytic/monoblastic, and megakaryoblastic leukemia (Supplementary Table S2) (14, 15). A “mixed lineage” leukemia was diagnosed if blasts were positive for myeloid- and lymphoid-associated antigens on flow cytometric analysis or expressed lymphoid antigens on flow cytometric analysis but had cytochemical staining reactions typical of AML.

Normality was assessed with a Shapiro–Wilk test when there was more than 3 data points per group. Non-Gaussian data was described as median and range, whereas 3 data points per group were described as mean and range. Medians of two groups were compared with a Wilcoxon signed-rank test (e.g., median percentage of anti-CD80-labeled monocytes vs. neutrophils). Proportions were compared with a Fisher’s exact test with a Bonferroni correction for the number of pairwise comparisons. Significance was set at p < 0.05.

CD80, as detected with the clone in this study, was consistently expressed on >95% of cells gated as monocytes (CD14⁺) and neutrophils (CD14⁻) in blood samples taken from 4 different healthy dogs. The intensity of CD80 expression (MFI) in CD14⁺ monocytes and CD14⁻ neutrophils was similar (Table 3). In contrast, cells gated as lymphocytes in the CD80 vs. CD14 experiments or triple-labeled with CD80 and B and T cell markers, CD21 and CD5, respectively, were negative for CD80 (Figure 1, representative images from one dog for CD80 vs. CD14 and a different dog for CD80 vs. CD21 or CD5). Events gated as lymphocytes in the CD80 vs. CD14 experiments had a median CD80 MFI of 6 units (range, 1–9 units), which was not significantly different from the median CD80 MFI of 5 units (range, 4–7 units) for the isotype control in that gate (p = 0.750).

On flow cytometric sorting with CD80- and CD14-labeled cells, only three populations were evident: CD80⁺/CD14⁺, CD80⁺/CD14⁻, and CD80⁻/CD14⁻ cells. No cells were CD80⁻/CD14⁺, suggesting that all CD14⁺ monocytes in blood from healthy dogs are CD80⁺. Differential cell counts on modified Wright’s-stained smears showed that monocytes, neutrophils, and lymphocytes dominated in the CD80⁺/CD14⁺, CD80⁺/CD14⁻ and CD80⁻/CD14⁻ populations, respectively (Figure 2, Table 4). Eosinophils were identified in the CD80⁻/CD14⁻ cells, indicating they are negative for CD80, when using the anti-hamster antibody (Table 4). Since eosinophils are found in low concentrations in blood and cannot be distinguished from neutrophils based on scatter characteristics or antibodies used in flow cytometric panels, their lack of binding of the anti-CD80 antibody was only uncovered by cell sorting.

When monocytes, B cells, and T cells were isolated via magnetic bead labeling with the anti-CD14, -CD21 and -CD5 antibodies then labeled with the anti-CD80 antibody, only CD14⁺ monocytes were CD80⁺, corroborating the results of the multiple labeling and cell sorter experiments. The purity of the isolated cells was generally >80% (Figures 3A–C, Table 5). Contaminating neutrophils in the CD14-magnetic bead isolated cells were weakly CD14⁺, suggesting that neutrophils were activated by the procedure and either expressed CD14 or had bound CD14⁺-microparticles shed from activated monocytes (Figure 3A). Similar to monocytes, neutrophils isolated from the double-density gradient were CD80⁺ (Figure 3D) with a purity of 78% or higher (Table 5).

When BMMC were incubated with the anti-CD80 antibody, a single CD80⁺ population with high SSC was identified (Figure 4) and comprised a median of 66% (range, 61–71%) of the living cells (n = 4). Typically, neutrophils and monocytes have higher SSC than lymphocytes. There were two populations of CD80⁻ cells; (1) One had similar high SSC to the CD80⁺ cells and comprised a median of 16% (range 16–23%) of living cells, and (2) Another population with low side SSC, corresponding to cells with less complexity, such as lymphocytes (Figure 4), comprising a median of 13% (range 11–21%) of living cells. When differential cell counts were done on modified Wright’s-stained cytospin smears of these 3 flow cytometric-sorted populations, the CD80⁺ cells consisted of neutrophil precursors (bands to myelocytes, with rare progranulocytes) and monocytes (Figure 4C). In contrast, the CD80⁻/high SSC cells were mostly mature and immature eosinophils with fewer monocytes, large reactive lymphocytes, and plasma cells (Figure 4C). Rare progranulocytes and basophils were seen. Lymphocytes comprised the majority of the CD80⁻/low SSC cells, with fewer monocytes, and there were reactive lymphocytes in this population (Figure 4C). Segmented neutrophils were not seen in any fraction because these cells settle at the interface between the 1.077 and 1.119 density gradients. The BMMC sorting showed that the anti-CD80 antibody binds to mature and immature neutrophils and affirms that it does not bind to mature and immature eosinophils. In addition, the antibody may not detect plasma cells. It is difficult to determine whether the antibody labels progranulocytes and basophils, given that these cells were only seen in low numbers in the cytospin smears and progranulocytes were identified in both CD80⁺ and CD80⁻ high scatter fractions.

B cell neoplasms were identified in 37 dogs from blood (n = 7) or lymph node (n = 27), bone marrow (n = 2), or pleural fluid (n = 1) aspirates. The dogs were a median of 8 years old (range, 2–13 years) with 20 female (2 intact) and 17 male (2 intact) dogs. Breeds consisted of 17 mixed breed dogs, 5 German Shepherds, 3 Golden Retrievers, 2 Australian Shepherds, 2 Rottweilers, and one each of the following: Belgian Tervuren Shepherd, Bernese Mountain Dog, Bichon Frise, English Springer Spaniel, Jack Russell Terrier, Shih Tzu, Vizsla, and Yorkshire Terrier. Five dogs with a moderate to marked lymphocytosis (median, 60.4 × 10⁶/mL, range, 18.1–146.8 × 10⁹/L), consisting of small to intermediate lymphocytes, were diagnosed with B-CLL. The remaining 32 dogs were diagnosed with B cell lymphoma/leukemia from lymph node or bone marrow aspirates or blood samples, with 1 concurring histologic diagnosis on a lymph node biopsy. None of the tumor cells were labeled with the anti-CD80 antibody (Figure 5, Table 6, Supplementary Figure S1, Supplementary Table S3). Of the other myeloid antigens used in this study, tumor cells were CD11c⁺ in 1/13 dogs (8%). A few dogs had aberrant CD3⁺ (7/37, 19%) or CD5⁺ (2/37, 5%) tumor cells; however, in all of these cases, tumor cells were CD21⁺ and CD22⁺, supporting a B cell neoplasm. In 3 dogs, there were discordant CD21 and CD22 reactions, with CD21⁻/CD22⁺ (n = 2) or CD21⁺/CD22⁻ (n = 1) cells (Supplementary Table S3). A B cell lineage was confirmed for 1 dog with a CD21⁻/CD22⁺ lymphoma/leukemia in the bone marrow on the basis of a clonal B cell population on polymerase testing for antigen receptor rearrangements (PARR) and CD20⁺/CD3⁻ cells on immunocytochemical staining of bone marrow smears (Supplementary Table S3).

T cell neoplasms were identified in 35 dogs from blood (n = 17) or aspirates from lymph node (n = 11), bone marrow (n = 2), mediastinal or lung masses (n = 2), or pleural (n = 2) or peritoneal (n = 1) fluid. The dogs were a median of 6 years old (range, 8 months to 14 years) with 12 female (1 intact) and 23 male (5 intact) dogs. Breeds consisted of 9 mixed breed dogs, 6 Golden Retrievers, 3 Labrador Retrievers, 2 German Shepherds, 2 Shih Tzus, and one each of the following: Australian Shepherd, American Bulldog, Bassett Hound, Bernese Mountain Dog, Bloodhound, Boxer, Bull Mastiff, Doberman, English Bulldog, Giant Schnauzer, Mi-Ki, Pug, and Staffordshire Bull Terrier. Two dogs with a lymphocytosis of granular lymphocytes (85.3 and 89.7 × 10⁹/L) were diagnosed with CD8⁺ T-CLL. Another dog had a mild lymphocytosis of granular lymphocytes (5.5 × 10⁹/L), with an average of 39% granular lymphocytes in bone marrow. The dog (a Golden Retriever) also had a lymphocytosis (8.7 × 10⁹/L) of T cells that lacked cytoplasmic granules and were CD45⁻ on phenotyping, supporting a concurrent indolent T zone lymphocytosis. Given the bone marrow infiltrates of granular lymphocytes, this dog was placed in the CD8⁺ T cell lymphoma/leukemia category (Supplementary Table S4). Another 33 dogs were diagnosed with T lymphoma/leukemia from blood or aspirates of lymph node, bone marrow, mediastinal or lung masses, or body cavity fluid, with 2 corroborating histologic diagnosis. One dog had a clonal T cell population with PARR in a lymph node aspirate. In 2 dogs (a Bull Mastiff and a Shih Tzu), the cells were CD45⁻ and had cytologic features of an indolent T zone lymphoma (6). None of the tumor cells in dogs with T cell neoplasms were labeled with the anti-CD80 antibody (Figure 4, Table 6, Supplementary Table S4); however, the cells were CD11b⁺ (2/22, 9%) or CD11c⁺ (6/19, 32%) in low numbers of dogs. There was discordant CD3 or CD5 expression in a few cases, with CD3⁻/CD5⁺ (n = 4) or CD3⁺/CD5⁻ (n = 5) cells. Tumor cells in 2 dogs were CD3⁻/CD5⁻ but were CD8⁺/ TCRαβ⁺ on flow cytometric analysis in one dog or strongly CD3⁺ on immunocytochemical staining in the other dog (Supplementary Table S4), supporting a T cell origin.

Thirty nine dogs were diagnosed with AML based on flow cytometric expression of myeloid-associated antigens (n = 33) or positive cytochemical staining reactions combined with negative staining for T (CD3/CD5) or B (CD21/CD22) lymphoid-associated antigens on flow cytometric analysis (n = 6) (Table 6, Supplementary Table S5). The phenotyping was done on blood (n = 23) or aspirates of bone marrow (n = 10), lymph node (n = 5), or pleural fluid (n = 1). In venous blood samples, blasts constituted ≥20% of a differential count in 22 dogs (22/23, 96%). The single dog with 10% blasts in blood had ≥20% blasts on cytologic examination of a lymph node aspirate (flow cytometric analysis was not done on the lymph node). The dogs were a median of 7.5 years old (range, 2–14 years) with 11 neutered female and 28 male (5 intact) dogs. Breeds consisted of 14 Golden Retrievers, 6 mixed breed dogs, 5 Labrador Retrievers, 4 German Shepherds, 2 Pembroke Welsh Corgis, and one each of the following: Anatolian Shepherd, Bernadoodle, Bernese Mountain Dog, Bulldog, Cockapoo, Maltese, Rhodesian Ridgeback, and Soft-coated Wheaten Terrier. Tumor cells were labeled with the anti-CD80 antibody in 28 cases (72%), mostly myelomonocytic and monocytic/monoblastic variants (Figure 4, Table 6, Supplementary Figure S1, Supplementary Table S5). However, 2 acute megakaryoblastic leukemias had CD80⁺ cells (Supplementary Table S5). A higher proportion of dogs had CD80⁺ cells compared to CD4⁺ (36%, 14/39), CD11b⁺ (44%, 17/39), CD11c⁺ (46%, 16/35), CD14⁺ (38%, 15/39), or CD18⁺ (56%, 10/17) cells; however, the difference was only significant for CD80⁺ vs. CD4⁺ (p = 0.003) or CD14⁺ (p = 0.006) (Supplementary Table S5). A few individual dogs with AML expressed only one myeloid antigen (CD11b in one dog, CD11c in one dog, CD80 in 4 dogs), however most dogs expressed different combinations of more than one myeloid antigen (Supplemental Table S5).

Eleven dogs were diagnosed with “mixed lineage” leukemia based on flow cytometric expression of myeloid-associated markers (n = 5) or positive cytochemical staining reactions combined with positive staining for B (CD21 or CD22) or T (CD3 or 5) lymphoid-associated antigens on flow cytometric analysis (n = 6) (Table 6, Supplementary Table S6). The phenotyping was done on blood (n = 8) or aspirates of bone marrow (n = 1) or lymph node (n = 2). Venous blood in 10 dogs contained ≥20% blasts and 1 dog had ≥20% blasts in a lymph node aspirate (Supplementary Table S6). The dogs were a median of 8 years old (range, 3–10 years) with 6 female (1 intact) and 5 male (2 intact) dogs. Breeds consisted of 3 Labrador Retrievers and one each of the following: Bernese Mountain Dog, Cavalier King Charles Spaniel, Golden Retriever, Labradoodle, mixed breed, Rottweiler, Swiss Mountain Dog, and Yorkshire Terrier. The neoplastic cells were weakly CD5⁺ or CD3⁺ (n = 9) or CD22⁺ (n = 3); no cases were CD21⁺ and 1 case was weakly CD3⁺/CD5⁺/CD22⁺. In two cases with weak CD3⁺ tumor cells on flow cytometric analysis, the tumor cells were negative for CD3 on immunocytochemical staining of smears, suggesting a false positive reaction. Tumor cells were labeled with the anti-CD80 antibody in 4 cases (36%), including 1 case that lacked other myeloid-associated antigens on flow cytometric analysis (Table 6, Supplementary Table S6). CD80⁺ cells were present in similar percentages to CD4⁺ (27%, 3/11), CD11b⁺ (45%, 5/11), CD11c⁺ (33%, 3/9), CD14⁺ (27%, 3/11), and CD18 (50%, 1/2) (Supplementary Table S6).